The structures of oligosaccharides excreted by sheep with swainsonine toxicosis.

Eleven oligosaccharides were purified form the urine of sheep with swainsonine toxicosis induced by the feeding of Astragalus lentiginosus. Oligosaccharides were extracted by charcoal adsorption, chromatographed on Bio-Gel P-2, and partially fractionated by preparative-layer chromatography. Separation into individual compounds was completed by semi-preparative high pressure liquid chromatography. Structures were determined by a combination of high pressure liquid chromatography and exo- and endo- glycosidase action, methanolysis followed by gas-liquid chromatography, methylation analysis, and high resolution nuclear magnetic resonance spectroscopy. Two homologous series of oligosaccharides were identified: (a) alpha-D-Manp-(1----6)-beta-D-Manp-(1----4)-D-GlcpNAc, alpha-D-Manp(1----3)-[alpha-D-Manp-(1----6)]-beta-D-Manp+ ++-(1----4)-D-GlcpNAc, alpha-D-Manp-(1----2)-alpha-D-Manp(1----3)-[alpha-D-Manp+ ++-(1----6)]-beta-D-Manp-(1----4)-D-GlcpNAc, and alpha-D-Manp-(1----2)-alpha-D-Manp-(1----2)-alpha-D-Manp+ ++-(1----3)-[alpha- D-Manp-(1----6)]-beta-D-Manp-(1----4)-D-GlcpNAc (minor series); (b) alpha-D-Manp-(1----6)-beta-D-Manp-(1----4)-beta-D-GlcpNAc- (1----4)-D-GlcpNAc, alpha-D-Manp-(1----3)-[alpha-D-Manp-(1----6)]-beta-D-Manp -(1----4)-beta-D-GlcpNAc-(1----4)-D-GlcpNAc, alpha-D-Manp(1----3)-alpha-D-Manp-(1----6)-beta-D-Manp -(1----4)-beta-D-GlcpNAc- (1----4)-D-GlcpNAc, alpha-D-Manp-(1----6)-alpha-D-Manp-(1----6)-beta-D-Manp++ +-(1----4)-beta-D-GlcpNAc - (1----4)-D-GlcpNAc, alpha-D-Manp-(1----3)-alpha-D-Manp-(1----6)-[alpha-D-Manp -(1----3)]-beta-D- Manp-(1----4)-beta-D-GlcpNAc-(1----4)-D-GlcpNAc, alpha-D-Manp-(1----3)-[alpha-D-Manp-(1----6)]-alpha-D-Man p-(1----6)-beta-D- Manp-(1----4)-beta-D-GlcpNAc-(1----4)-D-GlcpNAc, and alpha-D-Manp-(1----3)-[alpha-D-Manp-(1----6)]-alpha-D-Man p-(1----6)- [alpha-D-Manp-(1----3)]-beta-D-Manp-(1----4)-beta-D-GlcpNAc- (1----4)-D- GlcpNAc (major series).

*This investigation was supported by Grants AM03564 and HD16942 from the National Institutes of Health, by Grants DMB 84-12590 and 85-45771 from the National Science Foundation, and by the Northeast Regional NSF-NMR Facility a t Yale University, which is funded by Grant CDP-7916210 from the Chemistry Division of the National Science Foundation. This publication is No. 1042 of the Robert W. Lovett Memorial Group for the Study of Diseases Causing Deformities, Harvard Medical School, and Massachusetts General Hospital, Boston, Massachusetts 02114. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed.
the genera Astragalus or Oxytropis (locoweeds) in the United States, or Swuinsona (Australia), is a major agricultural problem (1, 2). The principal symptoms of intoxication are neurological and resemble those of bovine a-mannosidosis (2, 3), which results from a genetic deficiency of lysosomal a-mannosidase activity. The toxic principle of locoweeds is swainsonine (together with the N-oxide) (4), a potent reversible inhibitor of certain a-mannosidase activities including the lysosomal enzyme. Swainsonine toxicosis can therefore be considered an induced a-mannosidosis, and poisoned animals would be expected to accumulate glycoprotein-derived "high mannose" oligosaccharides in tissues ( 5 ) and body fluids, and to excrete them in urine (6). Oligosaccharides of this type were required in our laboratory for the synthesis of dolichol intermediates (7, 8). For this reason, oligosaccharides were isolated from the urine of locoweed-poisoned sheep and the structures determined as reported in this paper and in a preliminary report (9). During the course of the study, the characterization of the excreted oligosaccharides was shown to be important for the early diagnosis of locoism and for the development of swainsonine toxicosis as a reversible animal model for a-mannosidosis (10, 11).

DISCUSSION
The determination of the structures of 11 oligosaccharides from the urine of sheep with swainsonine toxicosis (see Table  IV) allows a detailed comparison with the oligosaccharides of human (24,25) and bovine (13)' mannosidosis. It also provides important information regarding the pathogenicity of swainsonine, and on the pathways of degradation of mannosyl oligosaccharides by lysosomal a-D-mannosidase.
The initial step in the structure determination of the purified oligosaccharides was high resolution HPLC,4 using synthetic compounds (14-17) or oligosaccharides isolated from human (12) or bovine (13) mannosidosis urine' as the reference materials. This HPLC analysis was performed on a mixture of oligosaccharides with or without appropriate del Portions of this paper (including "Experimental Procedures," "Results," Tables 1-111, V, and VI, Figs. 1-4, and Footnotes 2 and 3) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press.
The abbreviations used are: HPLC (LC in Miniprint Section), high pressure liquid chromatography; TLC, thin-layer chromatography; PLC, preparative-layer chromatography; Endo D and Endo H, endo-P-N-acetylglucosaminidase D and H; TR, retention time; MS, mass spectrometry; GC, gas chromatography.  dFor definition of binding and conditions of chromatography on concanavalin A-Sepharose, see Table VI chromatography see "Experimental Procedures," Miniprint Section).
Digestions with Endo D and Endo H were performed as described in Ref. 9.
rivatization (18), and the next step was digestion of the mixture with Endo H or Endo D. Initially, reduction to the alditols was routinely performed prior to the digestions. Later it was discovered that at least one of the oligosaccharides, Man3GlcNAcp-Ib (for structures see Table IV) is unexpectedly resistant to the action of Endo H when it has been reduced to the alditol (32). Therefore, the reduction was performed after the incubation and prior to HPLC. By HPLC analysis of the digest, including the identification, where possible, of new peaks, preliminary structural data could be obtained for some of the compounds. As a result of these experiments, it became apparent that the urine oligosaccharides of swainsonine-intoxicated sheep, like those of bovine mannosidosis, consist of two homologous series, ( a ) a major series, with an intact di-N-acetylchitobiose residue (G, series), and ( b ) a minor series, with a single "reducing" terminal 2-acetamido-2-deoxy-D-glucose residue (G series). When individual oligosaccharides had been purified, compositional analyses of per(trimethy1)silylated methyl glycosides, following methanolysis, confirmed these findings and provided a molecular formula for each compound. In some cases, this was also confirmed by chemical-ionization mass spectrometry (9). Further analysis of the purified oligosaccharides again involved high resolution HPLC, combined, in the case of the Gz series, with Endo D and Endo H digestions. It is noteworthy that, knowing the substrate requirements of these glycosidases (26,27), this combination of techniques was a convenient and rapid way of obtaining a provisional structure for the compounds, because the products of the digestions could often be identified by HPLC comparison with appropriate "standards." The potential pitfall of this approach is the reliability of HPLC separations (see comments below), but this problem has been largely resolved by modern column-packing technology, resulting in superior performance, and by performing the HPLC after perbenzoylation (18). The other analytical techniques employed in this study were chemical ionization mass spectrometry of permethylated alditol acetates (22) (methylation analysis) and high resolution (500 MHz) 'H NMR spectroscopy. In the following discussion, individual oligosaccharides will be referred to by their abbreviated formula (see Table IV, which shows the structure determined for each compound). The presence of a "nonreducing" terminal a(14)-linked D-mannose residue in trisaccharide Man,GlcNAc, tetrasaccharide ManpGlcNAcp, and in the corresponding compounds from bovine mannosidosis urine (13); instead of a (1+3) linkage as found in the smallest oligosaccharide of human mannosidosis urine (cr-D-Manp-(1+3)-P-D-Manp-(1~)-0-GlcpNAc) (24, 25), suggested a different substrate specificity for the residual a-mannosidase activity in humans and ru-minants. By studying swainsonine-induced mannosidosis in human fibroblasts, Cenci Di Bello et al. (33) concluded that a residual a-mannosidase activity, specific for a( 1 4 ) linkages, is inhibited by swainsonine in human cells, because the swainsonine-treated normal and mannosidosis cells both accumulated Man3GlcNAc instead of the usual trisaccharide. The lack of removal of the last a-linked mannose residue in most types of genetic and induced a-mannosidosis suggests that compounds containing this single a-linked residue are very poor substrates for lysosomal a-mannosidases.
The presence of an intact di-N-acetylchitobiose moiety in ManzGlcNAcz, Man3GlcNAcp, ManrGlcNAcz, and Man5GlcNAcz, and the corresponding oligosaccharides of bovine mannosidosis (13),' suggests a fundamental difference in the catabolic pathway of oligosaccharide catabolism in humans and other mammals. This is supported by studies on amannosidosis in cats (31, 34), @-mannosidosis in goats (35), swainsonine toxicosis in pigs (36), and GM1 gangliosidosis in dogs (37) and cats (38). The difference could be due to a deficiency or inhibition of an endo-8-N-acetylglucosaminidase activity (39) in non-human species, or an enhanced activity in these species of a peptide N-glycanase (39). The recently reported absence of an endo-@-N-acetylglucosaminidase activity in kidney of sheep, pigs, and cattle supports the former conclusion (40). Furthermore, ovine and bovine tissues apparently contain an endo-8-N-acetylglucosaminidase activity with a different substrate specificity (see comments below). Such an activity could be responsible for the production of tetrasaccharide Man3GlcNAc from pentasaccharide Man3GlcNAcz-Ia. ManaGlcNAc is a minor component of human mannosidosis urine (24, 25), the major tetrasaccharide having a completely different structure, again presumably reflecting differences in a-mannosidase specificity or (less likely) different substrate glycoproteins in humans.
The three pentasaccharides of the GZ series had closely similar retention times in HPLC.
Two of them, Man3GlcNAcz-Ia and Man3GlcNAcz-11, could not be separated by normal phase chromatography but could be separated by reversed phase chromatography after perbenzoylation. For this reason, compositional data were obtained only for ManaGlcNAcz-Ib and Man3GlcNAcz-11, the major isomers. On the other hand, isomers Man3GlcNAcz-Ia and Man3GlcNAcz-Ib could be separated on a preparative reversed phase column after perbenzoylation. The branched pentasaccharide Man3GlcNAcz-Ia is presumably derived mainly from the trimannosyl "core" of complex chains (41). Thus, it is more prevalent near the start of swainsonine ingestion (11), resulting from the catabolism of chains formed before toxicosis, and becomes a minor component as toxicosis proceeds, when the majority of chains presumably become "hybrid" owing to swainsonine inhibition of Golgi mannosidase I1 (42, 43) (see comments below). For similar reasons, it is the major pentasaccharide of bovine mannosidosis urine, because genetic mannosidosis does not involve aberrant processing. The origin of the linear pentasaccharides Man3GlcNAcn-Ib and Man3GlcNAcz-I1 is presumed to be selective hydrolysis of either the a(1-6) or a(1-3) linkages of hexasaccharide Man4GlcNAcz-I1 by residual lysosomal a-mannosidase activity (33).
The structures of pentasaccharide Man4GlcNAc and hexasaccharide MansGlcNAc, which are also present in bovine mannosidosis urine (13),' are surprising because they are not related to the structures of Man4GlcNAcz or Man5GlcNAcz. Instead, the structures are consistent with their formation by the "alternative" glycosylation pathway that involves a Man5GlcNAc2 dolichol intermediate (44), and their subse-quent release from either lipid-linked oligosaccharide or glycoprotein by an endo-/3-N-acetylglucosaminidase that recognizes structures with a trimannosyl core and an unsubstituted a(l-&)-linked D-mannose residue. Therefore, bovine and ovine tissues may not only lack an endoglycosidase that human tissues have but may possess an alternative enzyme, as mentioned earlier.
Two other hexasaccharides were identified, both having an intact chitobiosyl residue. In earlier work (91, a compound, Man4GlcNAcz-I, was isolated and characterized by digestion with Endo D and Endo H, resistance to /3-N-acetylglucosaminidase, and digestion by a-mannosidase. Its structure was determined to be that shown in Table IV. This is now known to be a minor isomer, its isolation being due to the limited resolving capabilities of the analytical and preparative HPLC columns employed in the initial phase of this work. The major isomer, Man4GlcNAcz-I1 has the structure shown in Table  IV. This structure was confirmed by 600-MHz two-dimensional J-correlated 'H NMR spectroscopy (collaboration with A. A. Bothner-By and R. L. Stephens, Carnegie-Mellon University, Pittsburgh, PA). The hexasaccharides Man4GlcNAcz-I and Man4GlcNAcz-II could be separated by chromatography on concanavalin A-Sepharose (32), and the result of this was consistent with their assigned structures. It also allowed a calculation of the relative proportions as approximately 1:9, Man4GlcNAcz-I to Man4GlcNAcz-II. This was in reasonably good agreement with the ratio obtained by perbenzoylation and reversed phase HPLC. These isomers presumably represent minor and major pathways of degradation of Man5GlcNAcz by residual lysosomal a-mannosidase activity (33), the major pathway involving initial cleavage of the a(1-3)-linked D-mannOSe residue of the trimannosyl core.
The heptasaccharide Man5GlcNAcz was examined by the same procedures as those employed for Man4GlcNAcZ-I, and shown to have the double-branched structure shown in Table  IV. Therefore, it can be concluded that all the major oligosaccharides from locoweed-intoxicated sheep urine have the same structures as those found in bovine a-mannosidosis (13).' However, it should be emphasized that the distribution of oligosaccharides into species with 2-5 mannose residues is completely different for the swainsonine-induced and genetic conditions. This is not a species difference: but reflects the dual role of swainsonine as an inhibitor of both lysosomal amannosidase and Golgi mannosidase I1 (42). Thus, swainsonine inhibits the processing of newly formed N-glycoprotein saccharide chains at the Man5GlcNAcz stage (44) and causes the formation of hybrid chains in cultured fibroblasts (43) or hepatocytes (45). If similar hybrid chains are assumed to be formed in swainsonine-fed animals, they would undergo normal catabolism in the presence of lysosomal fucosidase, Nacetylneuraminidase, P-galactosidase, and N-acetylglucosaminidase until Man5GlcNAc2 is formed. The conversion of MansGlcNAcz to Man4GlcNAcz-11, which becomes the major excreted compound in fully established toxicosis ( l l ) , is presumably the result of either residual lysosomal a-mannosidase activity or of the activity of another mannosidase not totally inhibited by swainsonine. The accumulation and excretion of high levels of MansGlcNAcz and Man4GlcNAcz contrasts to the situation in bovine mannosidosis, where Man5GlcNAcz and Man4GlcNAcz are relatively minor components, being derived from "high mannose" chains. On the other hand, the identical nature of the oligosaccharide structures in the induced ovine and genetic bovine mannosidosis conditions, and the similarity of clinical and pathological changes that occur in affected animals, suggests that deficiency of the lysosomal P. F. Daniel, and C. D. Warren, unpublished results. degradation of glycoprotein oligosaccharides is the primary cause of swainsonine toxicosis (6). It is interesting to note that glyco-asparagines, having the same structures as ManSGlcNAcp, Man4GlcNAcn-11, and ManZGlcNAcz, were isolated from the urine of patients with Gaucher's disease (46). Their accumulation was said to arise from an obstruction of lysosomal function by gross storage of glucocerebroside.
The comparison of the relative abundance of individual oligosaccharides in urine pooled from sheep that had ingested locoweed over a 7-week period with that in urine collected when swainsonine toxicosis was fully established (a single urine sample collected 6 weeks after the start of locoweed feeding) allowed some conclusions to be drawn concerning how the animals adapt to chronic intoxication. Thus, after long-term treatment it appears that sheep attempt to compensate for loss of normal lysosomal a-mannosidase activity by synthesis of an a-mannosidase that is only partially inhibited by swainsonine. Hence the relative abundance of ManSGlcNAcz is greatly decreased in the long-term treatment, and that of Man2GlcNAca and MansGlcNAc2 is increased. The proportion of "G series" oligosaccharides (e.g. Man,GlcNAc) is also significantly increased from 5 to 15%, indicating increased endo-8-N-acetylglucosaminidase activity or increased flux through the alternate glycosylation pathway.
A preliminary investigation of the oligosaccharides accumulated in the tissues of rams and pregnant ewes showed that each tissue is characterized by a unique pattern of oligosaccharides and that fetal and adult tissues differ with regard to the pattern and level of stored oligosaccharides.6 The continuation of this study, and correlation of the results with tissue a-mannosidase levels (Ref. 36, collaboration with 0. Touster and D. R. P. Tulsiani, Vanderbilt University, Nashville, TN) is in progress. Finally, it is important to note that the high mannose oligosaccharides described here are excellent reference compounds for studies of N-glycoprotein saccharide chain structure and biosynthesis (42, 43).